Originally posted by: EBKN_Eduardo_Aime

 Hello,

Actually (v12.5.1),  can I set a starting solution for a CP model in OPL Script?

In other words, can be made a CP version of the WarmStart.mod example?

Thank you very much for your help

#CPOptimizer
#DecisionOptimization

Originally posted by: Philippe_Refalo

With CP Optimizer providing a starting point will look like this:

You do not need to provide a value for every variables, in this case the assignments will be used as an indication for finding a first feasible solution. 

Regards

Philippe

#CPOptimizer
#DecisionOptimization

Originally posted by: EBKN_Eduardo_Aime

Thanks a lot Philippe.

 

I tried the following:

The model is syntactically correct.

Runs well the first solution, but in the second run indefinitely.

I'm not sure how to do the step: "fill solution".

 

/*********************************************

 * OPL 12.5.1.0 Model

 *********************************************/

using CP;

 

ranger = 1..10;

dvar int+x[r];

dvar int+y[r];

// The following array of values (defined as data) will be used as

// a starting solution to warm-start the CP search.

int RestartValues[iin r] = (i==5)? 10 : 0;

 

minimize

  sum(i in r ) x[i] + sum( j in r ) y[j];

subject to{

  ctSum:   

    sum( i in r ) x[i] >= 10;

  forall( j in r )

    ctEqual:

      y[j] ==j;

}

 

main{

  thisOplModel.generate();

  var ThisModelDef = thisOplModel.modelDefinition;

  /* Default behaviour */

  writeln("Default Behaviour");

  var cp1 = new IloCP();

  var opl1 = new IloOplModel(ThisModelDef, cp1);

  opl1.generate();

  cp1.solve();    

  writeln(opl1.printSolution());  /* Solution OK */

 

 /* Setting initial solution  */

  writeln("Setting Initial Solution ");

  var cp2 = new IloCP();

  var opl2 = new IloOplModel(ThisModelDef, cp2);

  opl2.generate();

 

  var RestartValues = thisOplModel.RestartValues ;

  var x = thisOplModel.x;

  var cpSolution = new IloOplCPSolution();

  for(var i = 1; i=10; i++)  

      cpSolution.setValue(x[i], RestartValues[i]);  /* Is this the right form to assign values ?? */

  cp2.SetStartingPoint(cpSolution);

 

  cp2.solve();  /* non-stop running !! */

  writeln(opl2.printSolution());

 

  opl1.end();

  cp1.end();

  opl2.end();

  cp2.end();

}

 

//OUTPUT:

//Default Behaviour

//// solution with objective 65

//x = [0 0 0 10 0 0 0 0 0 0];

//y = [1 2 3 4 5 6 7 8 9 10];

//

//Setting Initial Solution

#CPOptimizer
#DecisionOptimization

Originally posted by: Philippe_Refalo

There is no "non stop run" in CP Optimizer : the loop that populates the solution must contain i == 10 and not i = 0 and you need to use cp2.setStartingPoint(cpSolution); with a lower case s. 

In this simple case, CP Optimizer finds a solution quickly by itself and does not make use of the starting point. This happens sometimes when the model is simple. 

Philippe

#CPOptimizer
#DecisionOptimization

Originally posted by: EBKN_Eduardo_Aime

You are right Philippe
thank you very much

Eduardo.

#CPOptimizer
#DecisionOptimization